Intel® Distribution of OpenVINO™ Toolkit
Community support and discussions about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all things computer vision-related on Intel® platforms.

facial-landmarks-35-adas-0002 reference

Poca__Ramon
Beginner
378 Views

What are the reference 35 points of the facial-landmarks-35-adas-0002 network? Only its relative location is explained.

Is there any map/array with the normalized references used for training somewhere?

0 Kudos
3 Replies
JesusE_Intel
Moderator
378 Views

Hi Ramon,

Thanks for reaching out. The facial-landmarks-35-adas-0002 model is an Intel pre-trained model that is not opened sourced. All public information regarding this model can be found on the models description.

I apologize for the inconvenience. 

Regards,
Jesus

Poca__Ramon
Beginner
378 Views

Hi Jesus,

I just mean the 0-1 normalized landmark points used. 

One of the use cases for the model is facial alignment, so it's useful to have references.

In https://docs.openvinotoolkit.org/latest/_models_intel_face_reidentification_retail_0095_description_... there are the 5 keypoints for the landmarks-regression-retail-0009 model.

I was asking if the 35 keypoints are documented similarly somewhere.

 

Regards,

Vladimir_Dudnik
Employee
310 Views

Hi,

you may want to look at gaze_estimation_demo which can use the facial-landmarks-35-adas-0002 model to see how key points handled.

Regards,
  Vladimir

Reply